The association of premorbid psychological factors--stress, personality, social environment--and cancer incidence has often been reported as present and occasionally as absent. Reviews of such studies, however, have pointed to possible flaws in their experimental design, with resulting questions about their conclusions. In addition, almost none of the studies has taken into account certain factors having an important impact on outcomes. In these studies, each member of a cohort, presumed initially without cancer, will have produced data on psychological factors on entry. Data on premorbid psychological and relevant demographic and health factors have already been collected. Followup is in progress and the premorbid data will be related to cancer incidence and mortality when the latter data are available. Several hypotheses have been proposed.